Reading Data

The data was collected from Kaggle and is part of the dataset titled ‘Cancer Deaths by Country and Type (1990-2016)’, which provides comprehensive information on cancer mortality trends across different countries and cancer types over a 26-year period.

#Importing important libraries (All may not be used but these are generally imporant for this kind of data analysis)
library(tidyverse)

## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.4     ✔ readr     2.1.5
## ✔ forcats   1.0.0     ✔ stringr   1.5.1
## ✔ ggplot2   3.5.1     ✔ tibble    3.2.1
## ✔ lubridate 1.9.3     ✔ tidyr     1.3.1
## ✔ purrr     1.0.2     
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(tidyr)
library(ggplot2)    
library(dplyr)     
library(plotly)

## 
## Attaching package: 'plotly'
## 
## The following object is masked from 'package:ggplot2':
## 
##     last_plot
## 
## The following object is masked from 'package:stats':
## 
##     filter
## 
## The following object is masked from 'package:graphics':
## 
##     layout

library(stats)

# Reading in the .csv file from the raw GitHub URL.
cancer <- read.csv("https://raw.githubusercontent.com/rishavsofer/NYU_R_Project/main/Cancer%20Deaths%20by%20Country%20and%20Type%20Dataset.csv")

# Showing the top rows of the data for visualization.
head(cancer)

##       Country Code Year Liver.cancer Kidney.cancer Larynx.cancer Breast.cancer
## 1 Afghanistan  AFG 1990     243.6637      39.47049      109.3342      766.5354
## 2 Afghanistan  AFG 1991     261.2418      41.37602      117.3117      823.2339
## 3 Afghanistan  AFG 1992     284.4436      44.10631      128.0716      901.0221
## 4 Afghanistan  AFG 1993     313.1368      47.42485      141.4296      996.4328
## 5 Afghanistan  AFG 1994     343.2297      50.71095      155.7546     1097.8952
## 6 Afghanistan  AFG 1995     371.9291      53.81093      168.4818     1186.0278
##   Thyroid.cancer Stomach.cancer Bladder.cancer Uterine.cancer Ovarian.cancer
## 1       79.82017       923.4952       148.1392       108.1194       68.99470
## 2       85.11102       989.7096       156.9774       115.0684       73.51170
## 3       92.24060      1078.4590       168.9905       124.5071       79.58896
## 4      101.20673      1192.0645       184.3477       136.4029       86.60245
## 5      110.67992      1316.5057       200.2469       149.3039       93.47354
## 6      119.03708      1425.6272       214.5046       160.8058       99.79714
##   Cervical.cancer Prostate.cancer Pancreatic.cancer Esophageal.cancer
## 1        317.0996        130.3028          125.9848          189.3686
## 2        336.3375        139.5326          133.8216          200.2654
## 3        362.1939        151.5814          144.4897          215.1851
## 4        394.2766        167.7904          157.3959          234.1602
## 5        428.1947        184.7531          170.3166          254.3053
## 6        457.1709        200.9940          182.4932          271.9487
##   Testicular.cancer Nasopharynx.cancer Other.pharynx.cancer
## 1          27.49423           73.29566             40.96262
## 2          29.03400           78.31531             43.15965
## 3          31.24132           85.00778             46.18636
## 4          33.99369           92.96377             49.87100
## 5          36.90399          100.94951             53.62337
## 6          39.27382          107.83341             56.79263
##   Colon.and.rectum.cancer Non.melanoma.skin.cancer Lip.and.oral.cavity.cancer
## 1                442.1571                 26.44616                   53.59964
## 2                476.4320                 28.27527                   57.14889
## 3                521.9227                 30.71815                   61.87610
## 4                577.3300                 33.83544                   67.50486
## 5                635.4031                 37.10337                   73.17588
## 6                688.4817                 40.04284                   78.31936
##   Brain.and.nervous.system.cancer Tracheal..bronchus..and.lung.cancer.
## 1                        163.8691                             797.2657
## 2                        174.1832                             853.1264
## 3                        188.3823                             927.8128
## 4                        205.2504                            1017.9647
## 5                        222.3836                            1110.9972
## 6                        239.0922                            1195.7509
##   Gallbladder.and.biliary.tract.cancer Malignant.skin.melanoma  Leukemia
## 1                             125.9362                14.29398  727.7634
## 2                             133.7814                15.24105  766.0402
## 3                             144.2876                16.50883  820.9565
## 4                             157.4710                18.03851  891.1342
## 5                             171.3113                19.60664  965.2861
## 6                             183.7626                21.04178 1033.6420
##   Hodgkin.lymphoma Multiple.myeloma Other.cancers
## 1         191.3674         50.71944      294.8397
## 2         203.5096         54.31764      311.4691
## 3         220.2080         59.14420      334.5660
## 4         240.7188         64.93858      362.4867
## 5         262.3241         70.93035      390.3347
## 6         280.9813         76.48989      416.2341

Cleaning and Re-Organizing the Data

The data required re-naming columns for consistency and re-organization into long format for ease of analysis.

# Renaming columns for consistency
cancer_renamed <- cancer %>%
  rename(
    Country = "Country",
    Code = "Code",
    Year = "Year",
    Liver_Cancer = "Liver.cancer",
    Kidney_Cancer = "Kidney.cancer",
    Larynx_Cancer = "Larynx.cancer",
    Breast_Cancer = "Breast.cancer",
    Thyroid_Cancer = "Thyroid.cancer",
    Stomach_Cancer = "Stomach.cancer",
    Bladder_Cancer = "Bladder.cancer",
    Uterine_Cancer = "Uterine.cancer",
    Ovarian_Cancer = "Ovarian.cancer",
    Cervical_Cancer = "Cervical.cancer",
    Prostate_Cancer = "Prostate.cancer",
    Pancreatic_Cancer = "Pancreatic.cancer",
    Esophageal_Cancer = "Esophageal.cancer",
    Testicular_Cancer = "Testicular.cancer",
    Nasopharynx_Cancer = "Nasopharynx.cancer",
    Other_Phariynx_Cancer = "Other.pharynx.cancer",
    Colon_Rectum_Cancer = "Colon.and.rectum.cancer",
    Non_Melanoma_Skin_Cancer = "Non.melanoma.skin.cancer",
    Lip_Oral_Cavity_Cancer = "Lip.and.oral.cavity.cancer",
    Brain_Nervous_System_Cancer = "Brain.and.nervous.system.cancer",
    Tracheal_Bronchus_Lung_Cancer = "Tracheal..bronchus..and.lung.cancer.",
    Gallbladder_Biliary_Tract_Cancer = "Gallbladder.and.biliary.tract.cancer",
    Malignant_Skin_Melanoma = "Malignant.skin.melanoma",
    Leukemia = "Leukemia",
    Hodgkin_Lymphoma = "Hodgkin.lymphoma",
    Multiple_Myeloma = "Multiple.myeloma",
    Other_Cancers = "Other.cancers"
  )

# Previewing the renamed columns
colnames(cancer_renamed)

##  [1] "Country"                          "Code"                            
##  [3] "Year"                             "Liver_Cancer"                    
##  [5] "Kidney_Cancer"                    "Larynx_Cancer"                   
##  [7] "Breast_Cancer"                    "Thyroid_Cancer"                  
##  [9] "Stomach_Cancer"                   "Bladder_Cancer"                  
## [11] "Uterine_Cancer"                   "Ovarian_Cancer"                  
## [13] "Cervical_Cancer"                  "Prostate_Cancer"                 
## [15] "Pancreatic_Cancer"                "Esophageal_Cancer"               
## [17] "Testicular_Cancer"                "Nasopharynx_Cancer"              
## [19] "Other_Phariynx_Cancer"            "Colon_Rectum_Cancer"             
## [21] "Non_Melanoma_Skin_Cancer"         "Lip_Oral_Cavity_Cancer"          
## [23] "Brain_Nervous_System_Cancer"      "Tracheal_Bronchus_Lung_Cancer"   
## [25] "Gallbladder_Biliary_Tract_Cancer" "Malignant_Skin_Melanoma"         
## [27] "Leukemia"                         "Hodgkin_Lymphoma"                
## [29] "Multiple_Myeloma"                 "Other_Cancers"

# Reshaping the data to long format
cancer_long <- cancer_renamed %>%
  pivot_longer(cols = c(
    Liver_Cancer, Kidney_Cancer, Larynx_Cancer, Breast_Cancer, 
    Thyroid_Cancer, Stomach_Cancer, Bladder_Cancer, Uterine_Cancer, 
    Ovarian_Cancer, Cervical_Cancer, Prostate_Cancer, Pancreatic_Cancer, 
    Esophageal_Cancer, Testicular_Cancer, Nasopharynx_Cancer, 
    Other_Phariynx_Cancer, Colon_Rectum_Cancer, Non_Melanoma_Skin_Cancer, 
    Lip_Oral_Cavity_Cancer, Brain_Nervous_System_Cancer, 
    Tracheal_Bronchus_Lung_Cancer, Gallbladder_Biliary_Tract_Cancer, 
    Malignant_Skin_Melanoma, Leukemia, Hodgkin_Lymphoma, 
    Multiple_Myeloma, Other_Cancers),
    names_to = "Cancer_Type",
    values_to = "Deaths"
  )

# Previewing the long format data
head(cancer_long)

## # A tibble: 6 × 5
##   Country     Code   Year Cancer_Type    Deaths
##   <chr>       <chr> <int> <chr>           <dbl>
## 1 Afghanistan AFG    1990 Liver_Cancer    244. 
## 2 Afghanistan AFG    1990 Kidney_Cancer    39.5
## 3 Afghanistan AFG    1990 Larynx_Cancer   109. 
## 4 Afghanistan AFG    1990 Breast_Cancer   767. 
## 5 Afghanistan AFG    1990 Thyroid_Cancer   79.8
## 6 Afghanistan AFG    1990 Stomach_Cancer  923.

Checking for NA’s within the data

To check if there are any NA’s in case they needed to be dealt with for more efficient analysis

# Counting NAs in each column
na_count <- cancer_long %>%
  summarise(across(everything(), ~ sum(is.na(.))))

# Preview the result
na_count

## # A tibble: 1 × 5
##   Country  Code  Year Cancer_Type Deaths
##     <int> <int> <int>       <int>  <int>
## 1       0   756     0           0      0

NA’s were found in the ‘Code’ column which is a column that codes country names so a closer look is taken at that

# Filter the rows where 'Code' is NA
na_codes <- cancer_long %>%
  filter(is.na(Code))

# Preview the rows with NA in 'Code' column
head(na_codes)

## # A tibble: 6 × 5
##   Country   Code   Year Cancer_Type    Deaths
##   <chr>     <chr> <int> <chr>           <dbl>
## 1 Caribbean <NA>   1990 Liver_Cancer    1644.
## 2 Caribbean <NA>   1990 Kidney_Cancer    537.
## 3 Caribbean <NA>   1990 Larynx_Cancer    758.
## 4 Caribbean <NA>   1990 Breast_Cancer   2515.
## 5 Caribbean <NA>   1990 Thyroid_Cancer   127.
## 6 Caribbean <NA>   1990 Stomach_Cancer  2897.

summary(na_codes)

##    Country              Code                Year      Cancer_Type       
##  Length:756         Length:756         Min.   :1990   Length:756        
##  Class :character   Class :character   1st Qu.:1996   Class :character  
##  Mode  :character   Mode  :character   Median :2002   Mode  :character  
##                                        Mean   :2003                     
##                                        3rd Qu.:2009                     
##                                        Max.   :2016                     
##      Deaths         
##  Min.   :    37.92  
##  1st Qu.:  4863.98  
##  Median : 12739.30  
##  Mean   : 22361.27  
##  3rd Qu.: 21418.05  
##  Max.   :213062.16

Checking NA’s in Code for Correlated Country

Based on the results of the head(na_codes) prompt, we are checking what Country names are associated with NA’s

# Get a list of unique countries where 'Code' is NA
countries_with_na_code <- na_codes %>%
  distinct(Country) %>%
  pull(Country)

# Print the list of countries
print(countries_with_na_code)

## [1] "Caribbean"     "North America"

All of NA’s were associated with country names “Caribbean” and “North America”
It should be noted that neither of these are country names but are rather names of geo-political regions

Decision: Omit NAs as well as other regional categories and retain only country-specific data

#Omitting all NAs

cancer.clean <- na.omit(cancer_long)

#Specifying the regions to exclude
regions <- c("World", "East Asia", "Western Europe", "Africa","Eastern Europe","Southeast Asia","South Asia","Latin America and Caribbean","Sub-Saharan Africa")

# Filter the dataset
cancer.clean <- cancer.clean %>%
  filter(!Country %in% regions)

Analysis 1: Global Cancer Mortality Trends Over Time

Objective: Identify overall patterns in cancer deaths globally.
Approach: Calculate total cancer deaths per year and visualize the trend using a line chart.

# Calculating total deaths per year

total_deaths <- cancer.clean %>%
  group_by(Year) %>%
  summarize(Total_Deaths = sum(Deaths))

# Linear model
model <- lm(Total_Deaths ~ Year, data = total_deaths)

# Creating the plot with linear fit
ggplot(total_deaths, aes(x = Year, y = Total_Deaths)) +
  geom_line() +
  geom_smooth(method = "lm", se = FALSE) +  
  labs(title = "Global Cancer Deaths Over Time with Linear Fit",
       x = "Year",
       y = "Total Deaths")

## `geom_smooth()` using formula = 'y ~ x'

#Examining coefficients and statistical significance
summary(model)

## 
## Call:
## lm(formula = Total_Deaths ~ Year, data = total_deaths)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -96175 -63062  -1096  37965 178059 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -276876408    3383910  -81.82   <2e-16 ***
## Year            142565       1689   84.39   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 68370 on 25 degrees of freedom
## Multiple R-squared:  0.9965, Adjusted R-squared:  0.9964 
## F-statistic:  7121 on 1 and 25 DF,  p-value: < 2.2e-16

Impressive R-squared value of 0.9965

Analysis 2: Cancer Type-Specific Mortality Trends

Objective: Understand the changing landscape of cancer types.
Approach: Calculate the proportion of deaths for each cancer type over time, visualize the trends using stacked area charts and identify cancer types with increasing or decreasing mortality rates.

# To analyze cancer type-specific trends, we'll need to calculate the proportion of deaths for each cancer type per year

cancer_type_proportions <- cancer.clean %>%
  group_by(Year, Cancer_Type) %>%
  summarize(Total_Deaths = sum(Deaths)) %>%
  group_by(Year) %>%
  mutate(Proportion = Total_Deaths / sum(Total_Deaths))

## `summarise()` has grouped output by 'Year'. You can override using the
## `.groups` argument.

# A stacked area chart is suitable for visualizing proportions over time:


ggplot(cancer_type_proportions, aes(x = Year, y = Proportion, fill = Cancer_Type)) +
  geom_area() +
  labs(title = "Proportion of Cancer Deaths by Type Over Time",
       x = "Year",
       y = "Proportion of Deaths")

Closer Look: Look at only the top 5 cancer types

# Filtering by top cancer types

top_cancer_types <- cancer.clean %>%
  group_by(Cancer_Type) %>%
  summarize(Total_Deaths = sum(Deaths)) %>%
  arrange(desc(Total_Deaths)) %>%
  head(n = 5) 

top_cancer_data <- cancer.clean %>%
  filter(Cancer_Type %in% top_cancer_types$Cancer_Type) %>%
  group_by(Year, Cancer_Type) %>%
  summarize(Total_Deaths = sum(Deaths)) %>%
  group_by(Year) %>%
  mutate(Proportion = Total_Deaths / sum(Total_Deaths))

## `summarise()` has grouped output by 'Year'. You can override using the
## `.groups` argument.

ggplot(top_cancer_data, aes(x = Year, y = Proportion, fill = Cancer_Type)) +
  geom_area() +
  labs(title = "Proportion of Deaths for Top Cancer Types Over Time",
       x = "Year",
       y = "Proportion of Deaths")

Analysis 3: Cancer Mortality Disparities Between Countries

Objective: Explore differences in cancer outcomes across countries.
Approach: Identify countries with high cancer mortality burdens.

# Calculating total deaths per country
total_deaths <- cancer.clean %>%
  group_by(Country) %>%
  summarize(Total_Deaths = sum(Deaths))

# Ranking countries by total deaths
ranked_countries <- total_deaths %>%
  arrange(desc(Total_Deaths))

# Getting top and bottom 5 countries
top_5 <- head(ranked_countries, 5)

# Creating a ggplot object
p <- ggplot(top_5, aes(x = reorder(Country, Total_Deaths), y = Total_Deaths, fill = Country)) +
  geom_bar(stat = "identity") +
  coord_flip() +
  labs(title = "Top 5 Countries by Total Cancer Deaths",
       x = "Country",
       y = "Total Deaths") +
  theme(legend.position = "none")

# Converting to interactive plot
ggplotly(p)

#Seeing absolute numbers of total deaths by country
print(top_5)

## # A tibble: 5 × 2
##   Country       Total_Deaths
##   <chr>                <dbl>
## 1 China            53065391.
## 2 United States    15089092.
## 3 India            14805223.
## 4 Japan             8268955.
## 5 Russia            8069324.

It should be noted that the top 5 countries in the world by population (Earth Database, 2005) are: China, India, United States, Indonesia and Brazil

Closer Look: Look at the order of magnitude difference between the top 5 countries by total deaths

# Calculating order of magnitude for Total Deaths
top_5 <- top_5 %>%
  mutate(Magnitude = floor(log10(Total_Deaths)))

ggplot(top_5, aes(x = Country, y = Magnitude)) +
  geom_bar(stat = "identity") +
  labs(title = "Order of Magnitude of Total Deaths by Country",
       x = "Country",
       y = "Order of Magnitude")

print(top_5)

## # A tibble: 5 × 3
##   Country       Total_Deaths Magnitude
##   <chr>                <dbl>     <dbl>
## 1 China            53065391.         7
## 2 United States    15089092.         7
## 3 India            14805223.         7
## 4 Japan             8268955.         6
## 5 Russia            8069324.         6

Discussion and Future Directions

Discussion

There is a general increase in the number of deaths caused by cancer per year, with the increase being linear with an R value of 0.9965 and a p value of 2.2e-16.
The cancer types with the highest proportions of cancer deaths are breast, colon-rectum, liver, stomach and tracheal-bronchus lung cancer in increasing proportionality. While most of this cancer types remained the same or decreased in proportion, incidence of tracheal-bronchus lung cancer increased over time.
The top 5 countries for cancer deaths in the world in decreasing order were China, the United States, India, Japan and Russia with Japan and Russia having a lower order of magnitude (6 vs. 7) than the other countries.
- It should be noted that by world population (Earth Database, 2005), the 4th and 5th most populous countries are Indonesia and Brazil, so the higher incidence of cancer deaths in Japan and Russia could be significant.

Future Directions

The results of this project point to many future directions in terms of research, including:

Inquiry into why the increase in cancer moralities by year is almost linear and exploring possible causes
Investigating why the incidence tracheal-bronchus lung cancer moralities increased over time and if this increase was linked with geographical disparities in cancer mortality rates by conducting a more in-depth analysis of regional variations and identifying potential underlying factors.
Investigating the increased incidence of cancer moralities in Russia and Japan by looking at socio-economic data from all countries and comparing this with the mortality rate
Utilize advanced statistical methods such as machine learning to predict future cancer moralities for countries based on past data.

Analysis of Cancer Deaths by Country and Type

Purpose of Analysis